AI for Hydro Power Plants

The company
VERBUND is Austria's leading electricity company and one of Europe's largest hydropower electricity producers. VERBUND Hydro Power operates the group's hydro- and wind power portfolio efficiently and at the same time protects humans, nature and the climate.

The Challenge
VERBUND is seeking out start-ups and innovative companies with solutions for the analysis of telecommunicated measurement data and predicting the performance of dams.
Measurement systems for dams typically constantly monitor and telecommunicate outward parameters which influence the dam performance, such as the water level, air and concrete temperatures, as well as the dam's corresponding reaction or performance. Previous prognosis models for the performance of dams have been limited to few measurement values such as deformations, and are based on the a priori determination of correlations, which are adjusted to the measurement data by means of scaling factors and linear regressions.
The sought-for solutions include independent detection of connections between input data (and their chronological sequence) and measurement data at the dams, such as seepage amounts at fill dams or deformation measurements at concrete dams, and generating prognoses. It should be noted that the respective reactions may be immediate or delayed, as well as either reversible or irreversible.

Solutions which can clearly illustrate the recognised connections, and therefore make a physical interpretation or a comparison with existing physical approaches possible, will be given preference.
VERBUND's goal is to receive the most accurate prognoses possible with low fluctuation margins, so that it is possible to recognise deviations quickly. Through the independent detection of complex connections, prognoses can be made more easily, or applied to further measurement data, and previously used physical models and assumed connections can be reassessed.

Optional extension:
In the event of deviations from the predicted behaviour, it should be recognised as independently as possible which factor or which connection is responsible for the deviation, which would make it easier to identify the causes.
Real data can be made available to participants for working out their solutions – please request it from startups@verbund.com!

Application details can be found here: www.verbund.com/en-at/about-verbund/responsibility/energy2050/challenge